Scribd
Upload
6 views

Factor analysis report

The document discusses the results of a factor analysis conducted on a dataset of 2436 samples and 26 variables, highlighting the significance of chi-square value, p-value, and KMO values which indicate the dataset's suitability for factor analysis. It identifies six factors based on eigenvalues greater than one, justifying their selection using Kaiser’s criterion, and notes the presence of a sixth factor with low loading values. The author combines K-means clustering and PCA for further analysis and visualization of the data.

Uploaded by

nhatminh2122003
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
6 views

Factor analysis report

The document discusses the results of a factor analysis conducted on a dataset of 2436 samples and 26 variables, highlighting the significance of chi-square value, p-value, and KMO values which indicate the dataset's suitability for factor analysis. It identifies six factors based on eigenvalues greater than one, justifying their selection using Kaiser’s criterion, and notes the presence of a sixth factor with low loading values. The author combines K-means clustering and PCA for further analysis and visualization of the data.

Uploaded by

nhatminh2122003
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

21127355 Nhat Minh Do

Thursday, 20 March 2025

Practice 04: Factor Analysis

Multivariable Analysis

Assesment:
Preprocessing: 100%
Exploratory Factor Analysis: 100%
Evaluate Model Fit: 100%
Requests completion 100%
Overall: 100%, though I don’t really know.

Request 01: Explain the meaning of chi_square_value, p_value,


KMO values
Data shape after cleaning: (2436, 26)
Bartlett's Test: Chi-square = 18184.306307820785 , p-value = 0.0
KMO Test: KMO = 0.8483267027192372

Bartlett's Test (Chi-square value and p-value):

• Chi-square value (≈ 18184.31):


This statistic measures how much the observed correlation matrix
diverges from the identity matrix. A large chi-square value, as seen
here, indicates that the correlation matrix is signi cantly di erent from
an identity matrix—implying that the variables are correlated enough to
justify using factor analysis.

• p-value (0.0):
The p-value tests the null hypothesis that the variables are
uncorrelated (i.e., the correlation matrix is an identity matrix). A p-value
of 0.0 (typically reported as less than 0.001) strongly rejects the null
hypothesis. In this case, it con rms that there is su cient correlation
among variables to perform factor analysis.

Kaiser-Meyer-Olkin (KMO) Test:

• KMO value (≈ 0.848):


The KMO statistic measures sampling adequacy for factor analysis. It
assesses whether the partial correlations among variables are small.
Values range from 0 to 1, where:

1
fi
fi
ffi
ff
21127355 Nhat Minh Do
◦ 0.80 to 1.0 is considered meritorious or excellent,

◦ 0.70 to 0.80 is good,

◦ 0.60 to 0.70 is mediocre,

◦ Below 0.60 suggests the data might not be suitable for factor
analysis.

• With a KMO value of approximately 0.848, the dataset is very well-


suited for factor analysis.
Overall, these test results indicate that the dataset is factorable—there is
su cient inter-correlation among variables, and the sampling adequacy is
high, meaning that the underlying structure (factors) can be reliably extracted
from the data.

2
ffi
21127355 Nhat Minh Do

Request 02: Students explain the eigenvalues and base on that


eigenvalues choose the best number of factor to do the Factor
Analysis. Explain why you choose this number.

Given the Eigenvalues:


Eigenvalues: [5.13457985 2.75337527 2.14814212 1.85250623 1.54846254
1.11066151
0.98067695 0.82430872 0.79516217 0.71828982 0.68602566 0.67614851
0.65177562 0.62297416 0.59623084 0.56244599 0.54329886 0.51434031
0.49437154 0.48263949 0.44865495 0.42280589 0.40028481 0.38773757
0.38183723 0.2622634 ]
Consider the Scree plot of Eigenvalues of each factor:

3
21127355 Nhat Minh Do
We see that 6 factors has eigenvalues > 1, so we choose 6 factors for Factor
Analysis (Kaiser’s criterion). Furthermore, below is the loadings of each
factors based on type of questionnaires.

4
21127355 Nhat Minh Do
Request 03: Students look at the loadings table explain the
signi cant of each factor versus each property. If there are factor(s)
that has no “high loading” value, you can remove theseand perform
Factor Analysis again with the remain factor. Otherwise, explain the
Factor Variance Table.
To be honest at this point I don’t know what else to do, there is the 6th factor
which doesn’t have the “high loading” value. The above heatmap has already
shown that the loadings of each factor are in uenced by the type of
questionnaire. Therefore I went ahead and combined K-means clustering to
create 3 clusters of the data and used PCA on the remaining factors for
visualization. Below is the visualization using PCA.

5
fi
fl

You might also like